White
Paper
LED Driver Basics
INTRODUCTION
This brief provides basic information on the topics, and puts them in perspective in relation to JST drivers.
EFFICIENCY
Efficiency typically depends on choice of external components and switching frequency. The switching frequency multiplies losses associated with
energy delivered to and then lost from capacitances and stray inductances. Snubbers used in flyback transformer designs also add to frequency
dependent losses. RMS resistive losses arise based on steady state current flow through the components, most particularly the main switching FET,
the freewheeling diode, and the inductor.
EMI filter components add to the conductive steady state RMS loss. Each EMI filter inductor has a series resistance.
Because the components strongly determine the efficiency, and because the input voltage source and output LED string are necessarily predefined, then IMPROVING efficiency is mostly a power electronics game. The game to play is to choose the right components and switching
frequency.
Saving cost arises from using smaller components, as feasible with higher switching frequencies. However, higher switching frequencies can cause
additional difficulty in passing conducted EMI specifications, should EMI be a requirement.
•
JST controllers provide the flexibility to switch up to 500 kHz. As such, an engineer maintains freedom in power optimization.
Less obvious is the distinction between buck and buck-boost converters. The JST drivers can seamlessly control buck and buck-boost converters.
What is special about using buck converters in LED drive? With a buck converter, the LED output power can be reduced to 0. Thus 0 to 100 %
dimming is possible.
There is a misconception that buck regulators cannot work with dimmers, and pass EMI conducted emissions testing. This is false, inasmuch as JST
drivers DO pass.
Buck regulators are more efficient than buck-boost converters. For the same given input and output specifications, and for same size / cost, buck is
more efficienct. Achieving same efficiency with buck-boost requires more size/cost.
Also, buck converters provide much lower output current ripple, making it a better option for reliability of the output capacitor. A buck output
capacitor current ripple, being much smaller, will heat up less, and therefore have longer life.
Furthermore, the JST solution is one stage of power conversion. One stage is typically more efficient than two.
Compatibility with dimmers is enabled at no cost to efficiency at full load.
Summarizing
•
The JST solution is a one stage power converter. This has the advantage of using only one power switch in series between input and
output at any given time.
•
JST controller can work with buck power conversion, with an EMI filter, pass conducted emissions testing, and interoperate with all
dimmers.
•
Buck conversion features much smaller output current ripple, and so JST enables longer lifetime for less cost on the output capacitor.
•
Efficiency / cost / size is better with a buck, so long as the output voltage is reasonable (10 to 50 Volts).
•
Dimmer compatibility is ensured at no cost to efficiency at full load.
•
Simply, JST can equal or better competitive switching regulator offline LED drivers.
•
JST’s one stage switching regulator offline driver is typically better in efficiency than AC LED drivers (80 % at best).
JADE
SKY
TECHNOLOGIES
Confidential
1
White
Paper
LED Driver Basics
HIGH VOLTAGE
High voltage LED stacks require low current. If high voltage LED stacking is desired, it requires a buck-boost converter. Beyond 75 Volts, the buck
may struggle to provide power to the output with sufficient power factor. High voltage output is not a problem for JST. Buck-boost operation
works.
•
JST operates with high voltage LED stacks. Buck-boost conversion is available with the same chip.
•
JST can operate in boost converter mode as well if the LED stack voltage is greater than even the maximum AC line instantaneous voltage.
BUCK OR BUCK-BOOST
The following is a general guideline for non-isolated:
•
•
110 VAC 50 Volts or less
buck
> 50 Volts
buck-boost
> peak VAC
boost
220 VAC 100 Volts or less
buck
> 100 Volts
buck-boost
> peak VAC
boost
So even at 80 or 90 Volts, if the input is 220 VAC, then a buck is still likely feasible.
Buck is always preferable because
•
it has lower current ripple (better efficiency, smaller components)
•
it has a lower output current ripple (better reliability for electrolytic capacitor)
Buck-boost is fine if needed for higher output voltage.
2
Confidential
JADE
SKY
TECHNOLOGIES
White
Paper
LED Driver Basics
BUCK, BUCK-BOOST, BOOST, AND FLYBACK
Isolation may be desirable within an LED driver for safety. In the case of isolation, the flyback converter is the standard low cost single stage
conversion topology. The flyback transformer operation matches that of the buck-boost, except the storage element is the magnetizing inductance
of a transformer.
The following table compares the different single stage LED converters.
TABLE 1. CONVERTER TOPOLOGIES, PRIMARY AND SECONDARY
All converters are asynchronous (one diode and one switch). The boost converter is italicized because it does not apply in this application note.
Converter
Primary Circuit
Device
Secondary Circuit
Device
Input
Capacitor
Current
Ripple
Buck
NMOS, low side
Diode, picks up the High
control and gate
current when
drive, current in
NMOS turns off.
inductor increases.
Output
Capacitor
Current
Ripple
Highest
Reverse
Voltage
Across
NMOS
Highest
Reverse
Voltage
Across
Diode
Efficiency Cost, Relative
Relative to to Buck
Buck
When To
Use
Low
Input
Input
0%
VAC=110,
Vout < 50
0%
VAC=220,
Vout < 100
Primary inductor
current flows
through the load.
Buck-Boost
NMOS, low side
Diode, picks up the High
control and gate
current when
drive, current in
NMOS turns off.
inductor increases.
High
Input +
LED
Input +
LED
2-3 %
penalty
Load not
connected to
inductor.
Boost
0 % possible,
or 2-3 % if
resize for
better
efficiency
VAC=110,
Vout > 50
VAC=220,
Vout > 100
And
Vout <
VACmax
NMOS, low side
Diode, picks up the Low
control and gate
current when
drive, current in
NMOS turns off.
inductor increases.
High
Input
Input
0%
possible
0 % possible
Vout >
VACmax
High
Input +
N*LED
Input/N +
LED
3-5 %
penalty
5-10 %
increase
likely
Isolation
Required
Load not
connected to
inductor.
Flyback
NMOS, low side
control and gate
drive, current in
transformer
primary increases.
Load not
connected to
inductor.
JADE
SKY
TECHNOLOGIES
Diode, picks up
secondary
transformer
current when
primary side
NMOS turns off
High
Confidential
3
White
Paper
LED Driver Basics
POWER FACTOR
Power factor = 1 if the load is a resistor. Power factor control seeks to emulate a resistor but ensuring the input current follows the shape of the
input voltage.
JST’s controllers to this power factor correction. The PFC is elegant, simple, effective. The PFC is NOT unnecessarily complicated. As a result, the
power factor is typically 0.975 at full load. It is not 0.95, 0.9, or other lower values. It is high. The load is very much like a resistor.
The JST controller is one stage. In this way, power factor control is combined with current delivery to the LED. The capacitor at the LED provides
the necessary filtering to avoid 120 Hz or 60 Hz strobe light appearance.
Even more significant, high power factor is maintained even as the dimming angle changes. JST’s proprietary control algorithms enable this
significant performance advantage. Other solutions lose power factor quickly as the power decreases from full load. JST’s power factor remains
over 0.9.
Essentially, JST’s LED driver emulates an incandescent light bulb, which is a resistor. The power factor remains high for an incandescent bulb as
power decreases. JST provides the same loading response to the dimmer.
FIGURE 1. POWER FACTOR VERSUS DIMMING
•
JST power factor correction yields 0.975 or greater power factor at full load.
•
Consistent verification of this power factor at customers.
•
High power factor maintained over the entire range of dimming.
•
JST’s LED controller emulates an incandescent light bulb.
4
Confidential
JADE
SKY
TECHNOLOGIES
White
Paper
LED Driver Basics
TOTAL HARMONIC DISTORTION
JST’s buck controller THD sits around 20-25 % typically. Depending on how the solution is tweaked, it can be lowered a bit.
The THD for buck-boost conversion is less than 10 %, maybe even 5 %.
THD is the measure of how much the input current does NOT look like a sine wave. If it is a sine wave, THD = 0. If not, then THD goes up.
The THD of a buck suffers some because the input voltage is less than the LED voltage for some of the line cycle. When Vin < Vled, then no power
conversion occurs. So there are parts of the line cycle with 0 input current. These dead zones do not affect the power factor significantly, but they
increase THD.
Buck-boost converters, on the other hand, can deliver power even with very low input voltage. As a result, the input current can follow the voltage
sine wave shape for virtually the entire line cycle. THD can be very low.
•
JST operates with a buck
•
JST’s buck conversion can achieve 20-25 % THD, matching or exceeding competition
•
JST’s buck-boost conversion can achieve 5-10 % THD, matching or exceeding competition
If the output voltage is high (80 Volts, 220 VAC input), and the buck works, but THD is too high, then of course one can change to buck-boost at that
point.
FLICKER AND SHIMMER AND MORE!
Flicker is nasty flashing at high but visible frequency. This can occur at any power level. Flicker is ugly and obvious. When flicker occurs, one
suspects that the circuitry is having great difficulty arriving at some kind of stable operating condition.
Flicker is quite strong (i.e. more like flashing on/off). It is very much noticeable to everyone and is typically viewed as unacceptable. The instability
in current (and triac mis-firing) is very obvious on the oscilloscope measurements of input current, output current, and input voltage. The JST301
solution is designed to avoid this type of flicker on all dimmers, analog and digital.
Shimmer appears as subtle fluctuations of light, often at very low light levels. The lamp appears to “shimmer,” suggesting some kind of instability
or noisy operation.
Shimmer is a very weak variation in light output. It is more like very faint candlelight. Even on the oscilloscope, it is hard to detect. Shimmer
typically is detectable only at the very dimmest settings. Whether shimmer is acceptable is very subjective and depends on the observer. The
JST301 solution is designed also to minimize shimmer. However, sometimes there are tradeoffs, and it is possible to see a little bit of shimmer at
the dimmest settings (below which our competition typically operates). Typically a 9W or greater driver solution can be implemented to have
negligible shimmer. For less than 9W, it is more challenging, and shimmer can be observed in some cases, but again at very low light levels.
Line Cycle Ripple current will flow into the output of the JST single stage converter. In one extreme, the output capacitor is so large that the LED
carries only DC current. In the other extreme, the output capacitor only carries switching ripple current, but not the line cycle current, and so the
LED carries the entire 120 Hz ripple current. Typically the ripple is 120 Hz, but there may be 60 Hz content if the dimmer is asymmetrical.
If the output capacitor is not large enough, then the LED will accept ripple current rather than the capacitor. This is fine, as long as the LED ripple
current does not result in visible LED fluctuation. It appears like a strobe light. The strobe light is not obvious, like flicker. Yet the effect can be
real. The ripple current in the LED, relative to the average, determines how much of an effect this fast-action ripple has on the light.
This 60/120 Hz ripple is a natural consequence of a one-stage power factor corrected LED driver. It is not a fault. It is normal. In most cases, it is
small enough that the visual effect is negligible. Increasing the output capacitor redces LED line cycle ripple current, and thereby the strobe effect.
Error Behavior is gross enough that one basically things “it does not work.” For example, as light increases with a dimmer, suddenly the light turns
off and then cycles back up. The light cycles up, then off, then up, then off over and over. The cycle may be seconds long. As another example,
one plugs in a light bulb, moves the dimmer up, and then suddenly the light has a blinding strobe effect. These error behaviors are crazy enough
that one cannot possibly claim interoperation with the relevant dimmer.
These effects are especially problematic when interoperation is with digital dimmers.
Causes of Flicker
JADE
SKY
TECHNOLOGIES
Confidential
5
White
Paper
LED Driver Basics
1.
Insufficient power drawn from the LED controller is often the problem. A dimmer requires some minimum power after the TRIAC fires.
2.
Poor power electronics design, or poor power electronics control.
3.
Controller problems handling the entire dynamic range of the dimmer.
4.
Controllers having trouble figuring out what kind of dimmer is present.
5.
Controllers not providing enough loading, particularly for digital dimmers.
Causes of Shimmer
1.
Shimmer occurs when there is small-scale variability. One example of variability is in the switching behavior at the edges of the line cycle,
where the input voltage is close to the LED voltage, and where the input voltage is very low.
JST’s parts have been carefully and thoroughly designed to interoperate with dimmers. The staff are experienced, and have used their engineering
know-how to go very deep in studying the interaction with dimmers. The result is a solution that operates with any type of dimmer. The solution
is different from other competitive chipsets that struggle to find out exactly what kind of dimmer is present. The JST solution operates with any of
the dimmers, and as a result does not have the flicker and shimmer issues.
Also, the JST solution provides enough flexibility to handle different power levels and input voltage amplitudes. In cases in the lab, when a JST
board did in fact have flicker or shimmer, JST engineers were able to eliminate the effects by appropriate external component value change.
•
JST solutions interoperate with any type of dimmer.
•
Flicker and shimmer associated with controllers operating with a wrong dimmer type do not happen. There is no wrong dimmer type.
•
JST’s solution provides flexibility to adjust component values to eliminate shimmer and flicker for a given application requirement.
•
JST method to match incandescent bulb characteristics eases the compatibility burden. Incandescent bulbs do not flicker.
PATENT INFORMATION
The products and circuits illustrated herein may be covered by one or more U.S. and foreign patents, or potentially by U.S. and foreign patent
applications assigned to Jade Sky Technologies, Inc.
LIFE SUPPORT POLICY
Jade Sky Technologies products are not authorized for use as critical components in life support devices or systems without the express written
approval of the CEO of Jade Sky Technologies. As used herein:
1. A life support device or system is one which, (i) is intended for surgical implant into the body, or (ii) supports or sustains life, and (iii) whose
sfailure to perform, when properly used in accordance with instructions for use, can be reasonably expected to result in significant injury or
death to the user.
6
Confidential
JADE
SKY
TECHNOLOGIES
White
Paper
LED Driver Basics
2. A critical component is any component of a life support device or system whose failure to perform can be reasonably expected to cause the
failure of the life support device or system, or to affect its safety or effectiveness.
©2013, Jade Sky Technologies, Inc.
REVISION HISTORY
TABLE 2. REVISION HISTORY
Revision
1.0
1.1
Description
Initial Release
Added information about buck or buck-boost based on Vin and Vout, including implication on THD.
Added buck, buck-boost, boost, flyback table
JADE
SKY
TECHNOLOGIES
Confidential
Date
June 2013
July 2013
7